Back
The workshop will take place at the Max Planck Institute for Intelligent Systems on 13 - 14 July 2023, supported by ELLIS funding from the state of Baden-Württemberg. It is co-hosted by ELLIS, LAION, and MPI-IS.
Our goal is to discuss recent developments in foundation models in an informal setting.Since space is limited, we plan to live stream the talks.
With questions, feel free to reach out to Jenia Jitsev, Bernhard Schölkopf, or Christoph Schuhmann.
- All times in Central European Summer Time (CEST)
- Location: Lecture Hall N0.002, MPI-IS Tübingen
- You can access the slides of the presentations by clicking on the underlined titles
Time
Topic
Speaker
Format
9:00 - 09:15
Introduction to the workshop
Jenia Jitsev & Bernhard Schölkopf
In person
9:15 - 9:30
Open foundation models
Jenia Jitsev
9:30 - 9:45
Foundation multi-modal datasets
Christoph Schuhmann
9:45 - 10:15
From Paella to Würstchen: Recipes for Efficient and Simple Text to Image Generation
Dominic Rampas, Pablo Pernias
10:15 - 10:30
Discussion
10:30 - 11:00
Coffee break
11:00 - 11:15
How bias in LLM disrupts the Law: how to tackle discrimination in generative AI
Sandra Wachter
Online
11:15 - 11:45
Data Rivers? Re-balancing the data ecosystem that makes generative AI possible
Sylvie Delacroix
11:45 - 11:50
11:50 - 12:05
Overview of efforts in Japan on foundation models
Rio Yokota
12:05 - 12:20
Diffusion models for Cultural Heritage Restoration
Lucia Cipolina Kun
12:20 - 12:35
Beating Equal-Odds for Guided Video Synthesis with Foundation Models
Patrick Esser
12:35 - 12:45
12:45 - 12:55
Orthogonal Adaptation of Foundation Models
Weiyang Liu
12:55 - 13:10
Evaluating and improving the German language capabilities of open-source LLMs
Björn Plüster
13:10 - 13:15
13:15 - 14:00
Lunch
14:00 - 15:00
In-person Discussion / working groups
15:00 - 15:15
The Future of Generalist Agent
Aran Komatzusaki
15:15 - 15:25
Moûsai: Text-to-Audio with Long-Context Latent Diffusion
Zhijing Jin / Ojasv Kamal
15:25 - 15:55
Building multi lingual large open language models in EU Horizons HLPT
Sampo Pyysalo
15:55 - 16:10
16:10 - 16:20
CLadder: Assessing Causal Reasoning in Language Models
Luigi Gresele / Yuen Chen
16:20 - 16:30
Questioning the Survey Responses of LLMs
Ricardo Dominguez-Olmedo
16:30 - 17:00
17:00 - 17:15
Multimodal and SSL Embedding Spaces
Hilde Kühne
17:15 - 17:45
Scaling up language-vision learning with pre-trained components
Mehdi Cherti
17:45 - 17:55
17:55 - 18:25
DataComp: In search of the next generation of multimodal datasets
Ludwig Schmidt
18:25 - 18:35
LAION-Medical - towards a holistic approach to large-scale multimodal medical dataset generation
Robert Kaczmarczyk
18:35 - 18:45
18:45 - 19:45
Dinner
19:45 - 20:30
Panel: "Towards competitive Large AI in Europe"
Jenia Jitsev, Bernhard Schölkopf, Cecile Huet, Giambattista Parascandolo, Ludwig Schmidt, Sepp Hochreiter, Jess Montgomery
In person/online
10:00 – 11:00
Why This is (Not) the End of Research in Generative AI: Stable Diffusion & the Revolution in Visual Synthesis
Björn Ommer
Registration Closed
This event is no longer open for registration.
If you have any questions, feel free to contact us:
The following Zoom link was used to livestream the workshop: https://eu02web.zoom.us/j/64214595458?pwd=ZzgvL1p0dmdKRXRhOUtSSU5tVWRYQT09
More information